For example, the main effect condition (T0, ZT1,T2,T3) and location (Biceps， triceps）. After two-way repeated measures ANOVA, the interaction condition×location was not significant, but the main effect conditon and the main effect location are significant. In this case, can I performed one way repeated measures ANOVA for the main effect condition and location? If possible, can you suggest a paper about this explanation
Lack of significance is not a proof of the absence of an effect. Therefore, a "non-significant" interaction is by itself no carte blanché to interpret main effects as if there was no interaction.
You should instead check if the data is sufficiently inconsistent with any kind of relevant interactions. If this is the case, it may be savely ignored when analyzing the main effects.
If you don't know what effect size for the interaction may be (non-)relevant, you can simulate data and check at what size of the interaction you get problems with the interpretation of the main effects.
Dear Wei,
Usually we check for main effect and interaction both. Its not the pre-requisite that either interaction or main effect or both are necessarily significant.
Regards
Vishwanath
In my opinion, the non-significant of interactions does not have a simultaneous connection with the main effects. You can simply report meaningfully the main effects. Sometimes the opposite happens, the main effects are not meaningful, but their interaction is meaningful. In this context, put everything into good prism software. Your analysis is ANOVA's two-way model.
In agreement with previous comments: So, no, you don't re-fit the data with one-way anova. You keep the two-way model, and examine the effects that were significant (and interesting to you).
yes you can analyse the main effect with significance. You can find it on Steel and Torry or Gomez and Gomez or Cohran and Cox
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS